Available online at www.sciencedirect.

com
Proceedings
of the
Combustion
Institute
Proceedings of the Combustion Institute 32 (2009) 1075–1082
www.elsevier.com/locate/proci

Extinction of methane/air counterflow

partially premixed flames
Tomoya Wada a,*, Masahiko Mizomoto b,
Takeshi Yokomori b, Norbert Peters a
a
Institut für Technische Verbrennung, RWTH-Aachen, Tempelergraben 64, 52056 Aachen, Germany
b
Department of Mechanical Engineering, Keio University, 3-14-1 Hiyoshi, Kohoku-ku, Yokohama 223-8522, Japan

Abstract

Flame interactions of partially premixed flames (PPFs) were experimentally and numerically studied
using counterflow configurations with CH4/air mixtures under atmospheric condition. The main aims of
this study were to describe the flame interactions in terms of variations in global strain rate (K) and to
understand flame extinction. Two premixed-fuel streams were characterized by equivalence ratios (/lean
and /rich). Experiments were conducted under the following conditions: /lean + /rich = 2, 0.5 < /lean < 0.7
(i.e., 1.3 < /rich < 1.5), and several K values. To describe the interactions, twin flames (TFs), also examined
under corresponding /, were used as a criterion, because the / values were within the flammability limits.
Experimental results of PPFs showed a different behavior than those of TFs: first, three different luminous
layers were observed; second, thin single-layered flames, with a nearly constant flame width, were observed
near extinction for several combinations of /; third, PPFs could exist for significantly higher K than TFs.
Numerical results supported the experimental results and showed that two reaction paths,
CH4 + OH ? CH3 + H2O and CH3 + O ? CO + H2 + H, were remarkably intensified by flame interac-
tion of PPF near extinction; the significance of the reaction paths was determined. In addition, it was found
that excess CO production plays an important role in extinction of PPFs.
Ó 2009 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

Keywords: Partially premixed flame; Premixed flame; Flame stretch; Extinction; Triple flame

1. Introduction as the study of a lifted diffusion flame and the par-

tially mixed field caused by local extinction [1].
A partially premixed flame (PPF) is a domi- Previous studies were conducted using a round
nant phenomenon in combustion stability of real jet to describe the flame structure of a tribracial
combustors; therefore, it is one of the main flame and to obtain its propagation speed. These
focuses of combustion research. PPFs are studies strongly relate PPFs with lift-off height
observed in many studies of real combustors, such [2,3]. According to these experimental, numerical,
and theoretical studies, strong flame interaction
mainly occurs at the curved flame front, and heat
release should be taken into consideration in the
*
Corresponding author. Fax: +49 (0)241 80 92923. determination of lift-off height. Recently, a study
E-mail address: [email protected] (T. of the tribracial flame was conducted using a
Wada). round jet under microgravity conditions. This

1540-7489/$ - see front matter Ó 2009 The Combustion Institute. Published by Elsevier Inc. All rights reserved.
doi:10.1016/j.proci.2008.05.093
1076 T. Wada et al. / Proceedings of the Combustion Institute 32 (2009) 1075–1082

research demonstrated an effect of flow field, from the bottom inner nozzle, and the rich mix-
which would explain the difference between exper- ture flows from the upper inner nozzle. N2 flows
imental results and other results obtained under a from the outer nozzles as a shield gas. The equiv-
normal gravity condition [4]. alence ratio (/) was used to describe the fuel con-
In order to understand PPFs, experiments centrations. V denotes the cross-section averaged
have been conducted using counterflow burners. velocity. For counterflow PPFs, the premixed-fuel
The main advantage of using this type of bur- streams were characterized by / and V and
ner is that it creates the simplest flow field and expressed as /lean and Vlean (/rich and Vrich),
can avoid the effect of flame curvature. In addi- respectively. The flow fields are characterized by
tion, the gradient of fuel concentration is fixed the global strain rate K = (jVleanj + jVrichj)/L,
only by varying the condition of the two and velocity balanced conditions are used. K
streams. Double-layered, merged, and triple-lay- was varied by changing L during the experiments
ered flames (DLFs, MFs, and TLFs) can only while keeping V constant. For TFs, two identical
be differentiated from PPFs using a burner with premixed-fuel streams were used.
the simplest flow field, such as a counterflow To obtain counterflow PPFs, experiments were
burner. DLF has two reaction zones and con- conducted under the following conditions under
sists of a diffusion flame (DF) and a lean pre- atmospheric conditions: /lean + /rich = 2, each /
mixed flame (LPF)/rich premixed flame (RPF), was varied in intervals of 0.1 in the ranges
while MF has a single reaction zone and corre- 0.5 < /lean < 0.7, and 1.3 < /rich < 1.5. To obtain
sponds to a DF. TFs, experiments were conducted with the corre-
Previous studies using counterflow burners sponding / values of PPFs. Numerical calcula-
were conducted to simulate partial mixing tions were conducted using OPPDIFF [8] along
caused by local extinction. Previously, DFs were with a detailed mechanism (GRI-Mech 3.0 [9]).
widely used in real combustors, such as gas tur- Heat loss caused by radiation was not considered
bines and in power generation. Therefore, stud- [10]. The calculated laminar burning velocity
ies were conducted using DLFs to understand under the stoichiometric condition was 38.85
the extinction of DFs under highly turbulent cm/s.
conditions [5–7]. The main focus of these studies
was the transition of the flame structure from
DLF to MF. According to the results, two types 3. Results and discussion
of extinctions exist depending on fuel concentra-
tions. One type occurs when DLF becomes MF 3.1. Experimental results
near extinction. The other occurs when DLF
preserves its configuration until extinction. Figure 1 shows images of typical PPFs. Lean
These results clearly imply that fuel concentra- and rich mixtures (/lean = 0.6 and /rich = 1.4)
tion affects not only the flame structure but also flow from the bottom and top sides of the flames.
its extinction. The left, middle, and right images are obtained at
In the present study, PPFs were created using a K = 150, 250, and Kext s1, respectively. In the left
counterflow burner. Fuel was added to both and middle images, three different luminous layers
streams to obtain rich and lean mixtures. How- can be observed, and a typical TLF structure is
ever, the difference in fuel concentrations in the demonstrated [11]. Two thin layers located at
streams to obtain distinct LPF and RPF was the lean and rich sides are LPF and RPF, respec-
smaller than that in previous studies. In other tively. A weak luminous layer corresponding to
words, the fuel concentrations in the mixtures DF exists between these layers. On the other
are slightly lower/higher than the stoichiometric hand, only a single luminous layer can be
concentrations, and the gradient of fuel concen- observed at the extinction in the right image.
tration in the mixtures is relatively flat. Using this Three kinds of flames, LPF, RPF, and DF, merge
setup, premixed flames (PFs) were obtained, and to form a single visible layer at extinction, and a
the flame interaction and extinction caused by a strong flame interaction affects the entire flame
partially premixing field were studied. structure.

2. Experimental and numerical procedures

Two double-coaxial contoured nozzles were

used. The inner diameters of the nozzles were 12
and 16 mm, respectively. The nozzles were placed
on one another, and nozzle separations (L) were
varied from 10 to 12 mm. CH4 and air were used Fig. 1. Typical PPFs with /lean = 0.6 and /rich = 1.4.
as the fuel and oxidizer, and their mixture flowed (left) K = 150 s1, (middle) K = 250 s1, and (right)
from the inner nozzles. The lean mixture flows K = Kext s1.
 T. Wada et al. / Proceedings of the Combustion Institute 32 (2009) 1075–1082 1077

laser computed tomography [12,13], are in good

agreement with experimental data.
For the same /lean, dflame decreases as K
increases. On the other hand, dflame increases with
/lean from 0.5 to 0.7 for the same K except for
K = Kext s1. dflame is nearly constant at extinction
(critical dflame). Increasing /lean or decreasing /rich
increases the strained lean/rich premixed burning
velocity, which causes the strained lean/rich pre-
mixed flame to move away from the stagnation
plane (SP) [14]. However, constant critical dflame
values do not follow this trend.
Figure 4 shows K at extinction (Kext) of PPF
and TF using the same /lean to compare the
Fig. 2. Temperature profiles of PPF with /lean = 0.6 and extinctions with reference data. The Kext values
/rich = 1.4 at varied K. of TF are in good agreement with reference data
[15]. The lean premixed-fuel streams of PPF and
TF are identical. The comparison of Kext results
Figure 2 shows the variation of temperature of PPF and TF shows that Kext is proportional
profiles with PPF for /lean = 0.6 and /rich = 1.4, to /lean//, and that PPF can exist at much higher
for various strain rates. The distributed tempera- K than TF with constant critical dflame. dflame
ture profile in the low strain case (i.e., depends on the combination of / and K. If the
K = 150 s1) becomes a single peak in the high combinations of / change, a different K is
stretch one, especially before extinction. LPF required for extinction. Therefore, Kext depends
and RPF shown in Fig. 1 were independent. on the combination of / even if critical dflame is
Therefore, in the low strain case, the flatter part constant. This constant critical dflame of PPF at
of the distributed profile is caused by the temper- extinction would describe not only the flame inter-
ature difference between LPF and RPF, and reac- action in the counterflow configuration but also
tions caused by excess fuel and oxidizer in the the triple point of the lifted DF. As shown by
post-flame regime. On the other hand, these the results, LPF, RPF, and DF are merged into
results suggest a drastic change in the flame struc- a single layer at extinction, which simulates the
ture of PPF at extinction compared with the low condition near the triple point. Furthermore, var-
stretch case and that of TF. ied combinations of / (difference in fuel concen-
Figure 3 shows the variation in flame width tration in the stream based on the stoichiometric
(dflame), which is defined as the distance between condition) for PPF at constant critical dflame pro-
inflection points of luminosity at the unburned vide an extensively detailed and wide-ranged
side, with /lean [12]. In this figure, /lean = 0.5 understanding of the triple point, even if Kext is
denotes a combination with /rich = 1.5 because not constant.
of the condition /lean + /rich = 2. In case of Figure 5 shows a comparison between the PPF
K = 150 s1, the values of dflame compared with with /lean = 0.6 and /rich = 1.4 and the TF with
those obtained numerically (dflame,val), which were / = 0.6 at extinction. Figure 5a and b shows the
validated with experimental data obtained by

Fig. 4. Comparison of numerically computed and

experimentally determined results of global strain rate
Fig. 3. Flame width results of PPF versus equivalence at extinction (Kext) of PPFs and TFs, for several
ratio of lean mixture. equivalence ratios of lean mixture.
1078 T. Wada et al. / Proceedings of the Combustion Institute 32 (2009) 1075–1082

Fig. 5. Comparison of (left column) PPF with /lean = 0.6 and /rich = 1.4, and (right column) TF with / = 0.6 at
extinction: (a) and (b) species profiles, and (c) and (d) net production rate profiles.

species profiles, Fig. 5c and d shows net produc- to stagnation points. For PPF, the fuel is con-
tion rate profiles, and Fig. 6a and b shows integral sumed (the product is produced) at a narrower
net production rates of species. The integral is a region than that for TF. This result clearly shows
spatial integration between the two boundaries that flame interaction in the post-flame region
along the center axis, which demonstrates the bal- allows the reaction zones to approach each other.
ance of the production/consumption rate for each The entire reaction zone shifted to the lean side.
species in the system. The left and right columns For example, the peak position of CO2 produc-
show the results of PPF and TF, respectively. tion is located at the lean side in Fig. 5a. In addi-
As shown in Fig. 5, PPF and TF differ in dflame tion, the negative part of the CO net production
and the location of the reaction zone with respect rate is also located at the lean side in Fig. 5b.

Fig. 6. Comparison of integral net production rates at K = 150 and Kext s1: (a) PPF with /lean = 0.6 and /rich = 1.4,
and (b) TF with / = 0.6.
 T. Wada et al. / Proceedings of the Combustion Institute 32 (2009) 1075–1082 1079

These results imply that flame interaction does not Figure 7c and e shows the comparison of the
occur equally at the lean and rich sides of the peak values of the CH4 production rate for each
flame, and this nonequilibrium condition would elementary reaction, R11, R53, and R98, of PPF
play an important role in the extinction of PPF. and TF on the rich and lean sides, respectively.
Figure 6 shows the difference in the extinction In case of a peak at the lean side of the flames
mechanisms of PPF and TF. For TF at extinction, shown in Fig. 7c, the absolute values of the peaks
although CH4 consumption is nearly constant, the are not the same for PPF and TF; however, the
CO production rate is higher while the CO2 pro- ratio between the magnitudes of the peaks of the
duction rate is lower than those at K = 150 s1. reactions are nearly identical for PPF and TF.
This trend can be explained by the incomplete This result suggests that CH4 is consumed by
combustion near extinction. The residence time the same elementary reactions in TF and PPF.
of excess CO from the identical flames decreases On the other hand, the ratio between the magni-
as the flames approach SP (i.e., the velocity tudes of the peaks changes for the peaks at the
increases). The incomplete combustion inhibits rich side of the flames shown in Fig. 7e. In partic-
CH4 oxidation and induces extinction. On the ular, the peak value of R98 is significantly intensi-
other hand, for PPF at extinction, CH4 consump- fied for PPF. This result shows that the excess
tion and CO production rates are significantly oxidizer from the reaction zone at the lean side,
greater than those at K = 150 s1. In addition, OH, reaches the reaction zone at the rich side
CO2 production rate is also greater. This signifi- and affects the reaction. Figure 7d and f shows
cant difference from TF results from the gradient the comparison of the peak values of the CO pro-
in the post-flame regime. Excess fuel or oxidizer is duction rate for each elementary reaction, R166,
consumed with shortage of reactants from the R167, R168, and R284, at extinction on the rich
opposite side of the flame and is associated with and lean sides, respectively. The CO production
an increase in CH4 consumption at extinction. rate at the lean side shows the same trend as the
Therefore, the varied trends of CH4, CO, and CH4 production rate. At the lean side, CO is pro-
CO2 consumption/production rates are caused by duced following the same reaction path as TF and
flame interaction and are important for extinction. PPF. However, the excess oxidizer from the lean
In these figures, the amounts of CH3, CH2O, and side, O atom, reaches the reaction zone at the rich
HCO are negligibly smaller than those of CH4, side, and intensifies the reaction R284. This ele-
CO, and CO2. In other words, production and con- mentary reaction directly produces CO from
sumption of these species are well balanced. CH3 and by a bypass reaction of CO correspond-
ing to the main CH4 oxidation path. In PPF, more
3.2. CH4 and CO reactions CH4 reacts with OH than in TF and excess CO is
produced by the bypass reaction of CH3 with O,
In PPF, CH4 is mainly consumed through the although the CO2 formation reaction does not
following elementary reactions: change. As shown by the results, these two reac-
R11 : O þ CH4 ! OH þ CH3 ð1Þ tions, R98 and R284, are important to understand
flame interaction in PPF. Therefore, these reac-
R53 : H þ CH4 ! CH3 þ H2 ð2Þ tions also play an important role in analyzing
R98 : OH þ CH4 ! CH3 þ H2 O ð3Þ TLFs, where flame interaction is a dominant
factor.
Most of CO reacts with OH to form CO2, and
CO is mainly produced through the following ele-
mentary reactions: 3.3. Extinction mechanism of PPF
R166 : HCO þ H2 O ! H þ CO þ H2 O ð4Þ
Figure 8 shows CO molar and net CO pro-
R167 : HCO þ M ! H þ CO þ M ð5Þ duction rate profiles of PPF at extinction with
R168 : HCO þ O2 ! HO2 þ CO ð6Þ 0.5 < /lean < 0.7. In all cases, CO is produced
R284 : O þ CH3 ! H þ H2 þ CO ð7Þ at the rich side near the SP because of excess
O and OH from the lean side. CO diffuses from
Figure 7a and b shows the CH4 and CO pro- the rich to the lean side of the flame across SP.
duction rates profiles of each reaction, (1)–(7), After crossing SP, CO is consumed in the region
with PPF at extinction. Two peaks can be between the peaks under the tested condition at
observed at the lean and rich sides of the flame. nearly the same position for different combina-
Even though these peaks are completely separated tions of /. Figure 9 shows the comparison of
at low K, the two reaction zones are overlapped at the integral CO consumption rate in the nega-
extinction. This result shows that CH4 is con- tive region between two peaks in PPF and TF,
sumed (CO is produced) at the flame front and as shown in Fig. 7. In PPF, the integral value
the reaction paths differ significantly on the lean remains nearly constant with increase in /lean.
and rich sides. In addition, these two peaks over- On the other hand, the integral value in TF is
lapped only in a narrow region. proportional to /.
1080 T. Wada et al. / Proceedings of the Combustion Institute 32 (2009) 1075–1082

Fig. 7. Comparison of CH4 and CO production/consumption by elementary reactions in a PPF with /lean = 0.6 and
/rich = 1.4 and corresponding TFs with / = 0.6 and 1.4 at extinction: (a) and (b) CH4 and CO production/consumption
rate profiles of each reaction for PPF; (c) and (e) comparison of peak CH4 consumption rates profiles of reactions R11,
R53, and R98; and (d) and (f) comparison of peak CO production rates profiles of reactions R166, R167, R168, and
R284. Figure 7c and d represent results at the lean side of the PPFs and of lean TF, and Fig. 7e and f represents results at
the rich side of the PPFs and of rich TF.

For PPF, the total amount of fuel fed is con- within 0.2 mm of SP shown in Fig. 5. Therefore,
stant, which is the same as that in TF under the the influence of diffusion is much greater than that
stoichiometric condition. The entire reaction zone of convection. In addition, these balances explain
is shifted by the combination of / in PPF, and the the constant critical dflame values. If the difference
mass flux of CO at SP is varied as mentioned in / is small (e.g., /lean = 0.7 and /rich = 1.3), CO
above. However, the consumption rate of CO in diffusion is small in the post-flame region. There-
the post-flame region is also varied. Therefore, fore, PPF can exist until the diffusion and con-
the integral values of PPF, shown in Fig. 9, are sumption are balanced, which is achieved at
the maximum ones. Above this value, the balance large K values. On the other hand, if the difference
between CO diffusion from the rich side and CO in / is large, the balance is disrupted at small K
consumption at the lean side is disturbed, and values. However, the integral consumption rate
the entire flame extinguishes. The flame exists in the negative part between the two peaks is lar-
 T. Wada et al. / Proceedings of the Combustion Institute 32 (2009) 1075–1082 1081

premixed-fuel streams were fixed at /lean + /

rich = 2, and the experimental conditions were
0.5 < /lean < 0.7. The global strain rates (K) were
varied to obtain different flow field at velocity bal-
anced conditions. For TFs, two identical pre-
mixed-fuel streams were used.
The following conclusions were drawn:

 At low K values, three different luminous lay-

ers, corresponding to LPF, RPF, and DF were
observed, which represents a typical triple-lay-
ered flame (TLF) structure. On the other hand,
a single luminous layer is observed at
extinction.
 The flame width (dflame), which is defined as the
distance between the inflection points of lumi-
nosity at the unburned side of PPF, decreases
as K increases for the same /lean. However,
dflame values are nearly constant at extinction
(critical dflame) even if the combination of / is
varied.
 With an increase in K at a fixed combination of
/ in PPF, the flame interaction is intensified
and the flame structure is changed. According
to the analysis at extinction, excess oxidizers
from the reaction zone at the lean side, OH
and O, can reach the reaction zone at the rich
side and intensify elementary reactions,
OH + CH4 ? CH3 + H2O and O + CH3
Fig. 8. Comparison of PPF at extinction: (a) CO molar ? H + H2 + CO. The former reaction con-
profiles and (b) net CO production rate profiles.
sumes more CH4 than that without flame inter-
action and the latter is a bypass elementary
reaction that produces CO from CH3. These
two reactions are important to understand
flame interaction, and therefore, need to be
considered in the analysis for another configu-
ration, such as TLF.
 For PPF and TF with / = /lean, CO is con-
sumed near stagnation plane (SP). In the com-
parison of integral CO consumption rate
profiles of PPF and TF, the results of PPF are
higher than those of TF and the values are nearly
constant even if the combination of / is varied.
The extinction of PPF occurs in this region due
to excess CO. If the difference in / is small (e.g.,
/lean = 0.7 and /rich = 1.3), CO diffusion is
small in the post-flame region. Therefore, PPF
Fig. 9. Comparison of integral net CO production rate can exist until the diffusion and consumption
in post-flame region of PPF and TF at extinction. are balanced, which occurs at large K values.
In other words, larger K values are required to
approach SP. On the other hand, if the difference
ger than that in TF. Therefore, PPFs extinguish at in / is large, the balance is disrupted at small K
much higher K values with constant critical dflame. values. Therefore, it can be said that Kext is not
constant, but critical dflame is constant.

4. Conclusion

In order to understand flame interaction and References

extinction in a counterflow PPF, experiments
and numerical calculations using a CH4/air mix- [1] N. Peters, Turbulent Combustion, Cambridge Uni-
ture were conducted. The fuel concentrations of versity Press, Cambridge, England, 2000.
1082 T. Wada et al. / Proceedings of the Combustion Institute 32 (2009) 1075–1082

[2] T. Plessing, P. Terhoeven, N. Peters, M.S. Man- [9] G.P. Smith, D.M. Golden, M. Frenklach, et al.,
sour, Combust. Flame 115 (3) (1998) 335–353. http://www.me.berkeley.edu/gri_mech/.
[3] J. Boulanger, L. Vervisch, J. Reveillon, S. Ghosal, [10] Y. Ju, G. Masuya, F. Liu, H. Guo, K. Maruta, T.
Combust. Flame 134 (4) (2003) 355–368. Niioka, Proc. Combust. Inst. 27 (1998) 2551–2557.
[4] A.J. Lock, A.M. Briones, X. Qin, S.K. Aggarwal, [11] R.D. Lockett, B. Boulanger, S.C. Harding, D.A.
I.K. Puri, U. Hegde, Combust. Flame 143 (3) Greenhalgh, Combust. Flame 119 (1) (1999) 109–
(2005) 159–173. 120.
[5] K. Seshadri, I. Puri, N. Peters, Combust. Flame 61 [12] M. Hirota, T. Yokomori, K. Yasuda, Y. Nagai, M.
(2) (1985) 237–249. Mizomoto, G. Masuya, Proc. Combust. Inst. 31
[6] C.K. Law, D.L. Zhu, T.X. Li, S.H. Chung, J.S. (2007) 893–899.
Kim, Combust. Sci. Tech. 64 (1989) 199–232. [13] T. Yokomori, M. Mizomoto, Combust. Flame 135
[7] M.A. Tanoff, M.D. Smooke, R.J. Osborne, T.M. (4) (2003) 489–502.
Brown, R.W. Pitz, Proc. Combust. Inst. 26 (1996) [14] R. Seiser, L. Truett, K. Seshadri, Proc. Combust.
1121–1128. Inst. 29 (2002) 1551–1557.
[8] R.J. Kee, J.A. Miller, G.H. Evans, G. Dixon-Lewis, [15] C.K. Law, D.L. Zhu, G. Yu, Proc. Combust. Inst.
Proc. Combust. Inst. 22 (1988) 1479–1494. 21 (1986) 1419–1426.